A NEW WAKE VORTEX ENCOUNTER PREDICTION … · de encuentros con estelas turbulentas, a fin de...

A NEW WAKE VORTEX ENCOUNTER PREDICTION SYSTEM

BASED ON SPATIAL DATA STRUCTURES

Bachelor’s Degree in Aeronautical Management

Written by Laura Hernández Herrera

Directed by Dr. Sergio Ruiz Navarro

Sabadell, July 8th

2015

i

Títol del Treball Fi de Grau

A new Wake Vortex Encounter Prediction System based on Spatial Data Structures

Autor[a]: Laura Hernández Herrera

Data: 08 de Juliol de 2015

Tutor[a]/s[es]: Dr. Sergio Ruiz Navarro

Titulació: Grau en Gestió Aeronàutica

Paraules clau

Català: Trobada amb Estela Turbulenta, Trajectòria 4D, Estructures de Dades Espacials,

Granularitat, Gestió del Tràfic Aeri, Detecció de Conflictes, Distàncies de seguretat.

Castellà: Encuentro con Estela Turbulenta, Trayectoria 4D, Estructuras de Datos Espaciales,

Granularidad, Gestión del Tráfico Aéreo, Detección de Conflictos, Distancias de seguridad.

Anglès: Wake Vortex Encounter, 4D Trajectory, Spatial Data Structures, Granularity, Air Traffic

Management, Conflict Detection, Safety distances.

Resum del Treball Fi de Grau

Català: El projecte proposat es basa en el disseny i desenvolupament conceptual d’un sistema

predictor de trobades amb esteles turbulentes, per tal de garantir una futura gestió del tràfic aeri més

segura i eficient. Adaptant l’ús de les tecnologies existents per a la detecció de conflictes de manera

adequada, el sistema ha de ser capaç de contemplar tot l’espai aeri Europeu (tot l’espai d’estats), fent

possible la seva implementació a nivell computacional, en termes de memòria requerida.

Castellà: El proyecto propuesto se basa en el diseño y desarrollo conceptual de un sistema predictor

de encuentros con estelas turbulentas, a fin de garantizar una futura gestión del tráfico aéreo más

segura y eficiente. Adaptando el uso de las tecnologías existentes para la detección de conflictos de

manera adecuada, el sistema debe ser capaz de contemplar todo el espacio aéreo Europeo (todo el

espacio de estados), haciendo posible su implementación a nivel computacional, en términos de

memoria requerida.

Anglès: The proposed project is based on the conceptual design and development of a wake vortex

encounter predictor system, in order to ensure a safer and more efficient future air traffic

management. Adapting the use of current technologies to detect conflicts properly, the system should

be able to consider all European airspace (all state space), making possible its computational

implementation in terms of memory required.

ii

CONTENTS

1. INTRODUCTION ............................................................................................................. - 1 -

1.1 Background Information ........................................................................................... - 1 -

1.2 Research Objectives .................................................................................................. - 4 -

1.3 Research Methodology .............................................................................................. - 5 -

1.4 Structure of the document ......................................................................................... - 5 -

2. STATE OF THE ART....................................................................................................... - 6 -

2.1 SESAR and Trajectory Based Operations (TBO) ..................................................... - 6 -

2.2 Wake Vortex Phenomenon ...................................................................................... - 11 -

2.2.1 Origin and stochastic behaviour ...................................................................... - 11 -

2.2.2 Potential hazards and safety distances ............................................................. - 15 -

2.3 Spatial Data Structures (SDS) and their use in to tools for ATM ........................... - 17 -

2.3.1 SDS applications for ATM purposes ............................................................... - 20 -

2.3.1.1 Collision Avoidance .................................................................................... - 20 -

2.3.1.2 MTCD based on SDS .................................................................................. - 22 -

2.3.1.3 Strategic Deconfliction based on SDS ........................................................ - 23 -

2.3.2 SDS enhancements to reduce the memory and time ....................................... - 25 -

2.3.2.1 Relational SDS (RSDS) .............................................................................. - 25 -

2.3.2.2 Time-Space Data Structure (TSDS) ............................................................ - 28 -

2.3.2.3 RTSDS: Combination of RSDS and TSDS ................................................. - 31 -

3. WAKE VORTEX ENCOUNTER PREDICTION SYSTEM ......................................... - 33 -

3.1 High Risk Region Identificator (HRRI) .................................................................. - 36 -

3.2 WVE Risk Analysis ................................................................................................ - 40 -

4. EXPECTED ALGORITHM PERFORMANCE ............................................................. - 46 -

4.1 Most-likely Scenario ............................................................................................... - 47 -

4.2 Worst case scenario ................................................................................................. - 49 -

5. CONCLUSIONS ............................................................................................................. - 51 -

6. BIBLIOGRAPHY ........................................................................................................... - 53 -

iii

FIGURES

Figure 1: Potential hazardous wake encounters at en-route airspace ........................................ - 2 -

Figure 2: Risk of dangerous encounters if 3D Free-Route operations are applied.................... - 3 -

Figure 3: 4D Trajectory Dimensions ......................................................................................... - 8 -

Figure 4: Comparison between the optimal vertical profile and the traditional one ................. - 9 -

Figure 5: Wake vortex roll-up process .................................................................................... - 11 -

Figure 6: Plan view and lateral view of wake vortex .............................................................. - 12 -

Figure 7: Typical pattern of wake vortex in standard atmosphere .......................................... - 13 -

Figure 8: Two phase decay...................................................................................................... - 14 -

Figure 9: Wake Vortex Encounter .......................................................................................... - 16 -

Figure 10: SDS conceptual representation .............................................................................. - 17 -

Figure 11: SDS content example ............................................................................................. - 19 -

Figure 12: Representation of a pairwise algorithm ................................................................. - 20 -

Figure 13: Neighbours search to filter some of the pairwise comparisons ............................. - 21 -

Figure 14: Relational SDS architecture ................................................................................... - 26 -

Figure 15: TSDS conceptual representation ............................................................................ - 29 -

Figure 16: Neighbourhood (shadowed) defined by geometrical arguments ........................... - 30 -

Figure 17: Wake Vortex Encounter Prediction System's Architecture ................................... - 34 -

Figure 18: System's Flowchart ................................................................................................ - 35 -

Figure 19: Current Safety Standards ....................................................................................... - 36 -

Figure 20: Pre-process scenario .............................................................................................. - 37 -

Figure 21: Method to determine the affected quadrant ........................................................... - 38 -

Figure 22: Pre-process Output ................................................................................................. - 38 -

Figure 23: 2D High Risk Region............................................................................................. - 39 -

Figure 24: 3D High Risk Region (cube) ................................................................................. - 39 -

Figure 25: Wake Vortex's Discretization ................................................................................ - 40 -

Figure 26: WV's Enveloping Tube .......................................................................................... - 41 -

Figure 27: Circumference with symmetrical and antisymmetric segments ............................ - 42 -

Figure 28: Octant formed by joint pixels ................................................................................ - 42 -

Figure 29: Octant formed by disjoint pixels ............................................................................ - 43 -

Figure 30: Discrete Circumference ......................................................................................... - 43 -

Figure 31: Construction of discrete cylinder ........................................................................... - 44 -

Figure 32: Wake Vortex’s Discrete Model ............................................................................. - 44 -

Figure 33: State space of the HRR .......................................................................................... - 45 -

Introduction

- 1 -

1. INTRODUCTION

1.1 Background Information

Given the spectacular growth of air transport in recent years and considering future

forecasts, it is essential the consolidation of a more unified and efficient European air

traffic management.

The SESAR program (Single European Sky ATM Research) is the technological

support and the means of implementation of the European project SES (Single

European Sky). One of its most important goals is the operations management based on

trajectories, in order to obtain, the optimal trajectories through a collaborative decision

process (Collaborative Decision Making) to obtain conflict-free trajectories, considering

the interests of all stakeholders, and with enough time-ahead to ensure the highest

possible safety and efficiency.

Since forecasts show a growth factor of 150% of air traffic up to 2020, and therefore,

there will be an increased traffic condensation, it will be necessary overturn the efforts

not only to detect conflicts and collisions, but also to avoid the hazards of wake vortex

encounters (WVE). Although they rarely cause accidents with consequences of great

magnitude, it may cause the destabilization of the aircraft and injury the crew and its

passengers.

Actually, in TMA the dangers of wake vortex encounters are well known and airborne

separation minima defined by ICAO are specifically designed to minimize the

underlying hazards, especially near the runway (with the aircraft closer to the ground

and closer to their flight envelope limits). Nowadays wake vortex hazard is one of the

most limiting factors in airport capacity, thus many efforts have been done to find new

ways of re-categorizing aircraft according to their wake vortex (RECAT project is an

example) and new procedures such as Time-Based Separations (TBS) have been

designed considering wake vortex behaviour and duration to allow ATC (Air Traffic

Control) and flights being more precise with the synchronization of the landing traffic.

Introduction

- 2 -

In en-route operations the dangers have been traditionally considered negligible because

the probability of a wake vortex encounter has been up to now low and the severity of a

possible encounter has been often found not worrying. Nevertheless, several factors that

may cause the increase of both the probability of encounter and the severity are

changing in the context of the new ATM proposed by SESAR: for the next decades

more traffic density is expected, as well as more heterogeneity in the aircraft mix (there

is a tendency to bigger and smaller aircraft coexisting in the same airspace) and new

concepts of operations may be introduced such as the concept of Free-Routing, or the

potential introduction of continuous operations for all the phases of flight, referred

sometimes as 3D Free-Routing (which implies the relaxation of Flight Level separation

standards, with an expected gain in the capacity and efficiency performances of the

network).

Figure 1 illustrates an example of potential hazardous wake encounters at en-route

airspace with changes in aircraft mix even when the current safety standards are applied.

Figure 2 illustrates the risk of dangerous encounters if 3D Free-Route operations (i.e.

continuous operations) were introduced in the ATM. Note that the relaxation of the

flight level schemes might cause severe encounters even if aircraft current separation is

maintained and that these identified hazards could become even worse if the current

separation standards are reduced on behalf of the airspace capacity.

Figure 1: Potential hazardous wake encounters at en-route airspace

Introduction

- 3 -

Thus, in the context of the future ATM proposed by SESAR for 2020+ [21], i.e., a

Trajectory Based Operations (TBO) system, some new wake vortex separation rules and

new strategic separation methods for all phases of flight (in particular for en-route) will

be required.

In particular, new strategies have to be found in order to store information about wake

vortex, and consequently, to predict wake vortex encounters, achieving Conflict

Resolution (CR) systems safer and more efficient since the potential risks of this

phenomenon are taken into account and the responsiveness to uncertainties arising is

improved.

Therefore, the main objective of this project is to extend and enhance previous

technologies used in TMA to be used in ECAC, since SDS current technology has been

successful in MTCD (Mid-Term Conflict Detection) and SD (Stategic-deconfliction)

due to its capability to efficiently store and manage information about all European

state-space.

This project is based mainly in the fields of information management and computer

science, since what we want to achieve is a way to join these two fields in order to be

able to handle micro-level state-space information of the traffic and associated wake

vortices at entire European ATM level.

Figure 2: Risk of dangerous encounters if 3D Free-Route operations are applied

Introduction

- 4 -

1.2 Research Objectives

The main purpose of the project is to develop a system capable to predict encounters with

wake vortices generated by aircraft flying over the airspace, considering all the European

air traffic (all state space), and reducing the amount of memory necessary to store all the

information required.

However, in order to achieve this objective successfully, it has been necessary the

definition of specific sub-goals, which are presented below:

Obtain a wider knowledge of the nature and behaviour of wake vortices.

Know the derivative risks and consequences of wake vortex encounters during

flight.

Become familiar with current conflict detection systems based on spatial data

structures (SDS) and identify gaps they may present.

Design a mechanism able to provide information about the position and strength

of wake vortices, in a certain period of time, caused by those aircraft that fly over

the modelled space.

Adapt this mechanism to consider the entire state space, i.e. the entire European

air traffic, reducing the amount of memory needed to facilitate their

implementation.

Design a high level architecture that contains this mechanism, and therefore, is

able to provide information about the severity in the time-space of this

phenomenon to guarantee a more efficient global conflict resolution.

Introduction

- 5 -

1.3 Research Methodology

The research methodologies followed in order to achieve each of the objectives set out

in Section 1.2 are listed below:

Analyze the related literature and preceding projects documentation.

Identify the shortcomings in the systems presented in the previous researches.

Develop a brainstorming about potential solutions to implement.

Identify the strengths and the weaknesses of each one.

Select the best option to be included in the system’s development.

Perform calculations to obtain results.

Analyze the results in order to identify the improvements obtained with respect

to the current systems.

1.4 Structure of the document

The structure of the document is composed by the following chapters:

Chapter 2: includes information about SESAR’s objectives and innovations as well as

its influence in future ATM. A part from that, data about the origin and behavior of

wake vortex are presented, also explaining the derived potential hazards. Finally, SDS

architectures used in current conflict detection systems are deeper explained due to their

importance in this project.

Chapter 3: presents the extension of the current SDS technology, describing each one of

the functional modules that configures its operational structure.

Finally, the results and improvements obtained using the proposed high-level

architecture are shown in chapter 4 while the conclusions are presented in chapter 5.

State of the art - SESAR and Trajectory Based Operations (TBO)

- 6 -

2. STATE OF THE ART

The contents presented in this chapter are the following ones:

- SESAR’s objectives and future ATM to understand the environment in which

the project is contextualized.

- Wake vortex and potential hazards because it is important to know the WV’s

behaviour and dangers of encounters in order to extend the current technologies

in such way that the derived uncertainties of this phenomenon can be

contemplated.

- Previous conflict detection systems based on Spatial Data Structures (SDS),

because they are the technological starting point to conceptually develop the

proposed high-level architecture.

2.1 Single European Sky ATM Research (SESAR) and

Trajectory Based Operations (TBO)

The European ATM system currently handles approximately 28,000 commercial flights

daily, with peaks up to 33,000 flights, aside from other types of traffic. The forecast of

growth to 2020 points to a traffic growth of 150% [12] what is a challenging task in

reference to their management and the necessary infrastructure. This implies that should

undertake operational and technological developments in various fields in order to

optimize both the use of airport infrastructure and the airspace, to reduce the operating

costs, and to improve the safety standards by minimizing the environmental impact [12].

In this context the European Union launched the "Single European Sky" initiative that

involves the development and implementation of a common transport policy in order to

achieve an effective air traffic management system able to allow the safe and regular

operation of its services, so as to facilitate the free movement of goods, people and

services. These goals are achieved through a more rational use of airspace, increasing its

capacity and ensuring a high degree of safety throughout the European level airspace

organization.


- 7 -

Contrary to what happens in the US, Europe has a fragmented sky where the air traffic

management is conducted at a national level. This situation greatly complicates the

management of air traffic in our continent. The Europe's Single European Sky initiative

aims to solve this fragmentation, which is a major bottleneck for airspace capacity,

structuring the air navigation services at pan-European level instead of at national level.

The implementation of the "Single European Sky", which is based on the intensive use

of modernize technologies and higher degree of automation, will allow to,

progressively, have more direct flight trajectories, reduce flight times, achieve fuel

savings and thus reduce pollutant emissions into the atmosphere, while increasing

punctuality and reducing costs [12].

The answer to the challenge of increasing air traffic in Europe is growing in two

different ways, but closely related: first the seamless management of the airspace with

global perspective (Single European Sky) and, on the other hand, operational tools and

technologies necessary for its proper management (SESAR).

SESAR (Single European Sky ATM Research) is a joint project of the European air

transport community whose aim is the implementation in 2020 of a European ATM

network of high performance, born from the need to establish an integrated and shared

vision on the current state and evolution of air traffic management system. Its aim is to

address, through the implementation of new operational concepts, procedures and

technologies, the demand increasing and the requirements planned for the coming years.

The SESAR JU’s general objectives can be summarized as:

Manage an increasing volume of air traffic until triple the current.

Reduce by 10% the environmental impact of each flight.

Reduce by 50% the costs associated with the management of air traffic.

Multiply by 10 the security of air traffic.

The achievement of these goals will enable a number of benefits in economic terms:

a) Social benefits (by the year 2020). Contributions to European gross domestic

product (sum of direct, indirect and induced contributions) equivalent to 470.000

M € a year.


- 8 -

b) Operational Benefits quantified (cumulative figures between 2008 and 2020):

In cost-effectiveness: 8,000 M €.

In increasing capacity and reducing delays: 9,000 M €.

In fuel savings: 8,000 M €.

In service’s quality: 2,000 M €.

One of the most important challenges of SESAR with respect the current ATM, which

establishes the basis of this project and will be present in the system’s development, is

the introduction of the Trajectory Based Operations (TBOs), implying the use of 4D

gate-to-gate precision trajectories (trajectories defined in the 3 spatial dimensions

together with a time-stamp), also known as Business Trajectories (BT) in the SESAR’s

terminology for civil flights. The idea behind this concept is that both airlines and

providers of Air Navigation Services (ANSP) and airport operators come to an

agreement, in a collaborative way, in order to find the optimal door to door flight plan

for each of the scheduled flights [1].

During the negotiation process between controllers and airlines, the Business Trajectory

is called Shared Business Trajectory (SBT). In this process, the company will propose

the more efficient trajectory for them to cover a certain distance between two airports.

Then the ANSP accepts or denies the trajectory depending on the constraints of airspace

and airports. This negotiation process is repeated until a trajectory is found and accepted

by both agents (except in critical situations where there is no time for negotiation, in

which case will prevail the trajectory imposed by the ANSP) [5].

Once one SBT is accepted by both agents, it passes to be called Reference Business

Trajectory (RBT). From that moment, this trajectory is considered optimal, because it

takes into account the constraints of the air traffic controllers and airline preferences

(assuming the choice of the least cost trajectory).

The Business trajectory (BT) requires a precise definition in the 4 dimensions, i.e., in

the three spatial dimensions and in the time (see Fig. 3). For this reason his name is

often replaced by 4D trajectory.

Figure 3: 4D Trajectory Dimensions

Source: Ruiz, 2010, [5]


- 9 -

In the absence of airspace or traffic constraints, “an optimal trajectory” basically consists of

following the most direct route possible and maintaining a uniform acceleration during take-

off and up to the maximum desired altitude where the aircraft can start a smooth and

continuous descent up to the runway, also using a uniform maneuver without accelerations

until get into contact with soil.

In this way, by applying the optimum vertical profile it is achieved that the planes are less

time in the air (thus increasing the capacity of the sky), whereas progressive acceleration

allows a significant fuel savings and trip-time, as well as a significant decrease of the CO2

emitted [5].

Another important concept inside the SESAR context is the term called “Strategic de-

confliction”. This term encompasses those actions whose objective is to minimize the amount

of conflicts, appearing because of a loss of separation between two aircraft. These actions are

taken by Network Manager (NM) which are always executed when the takeoff time is

known with sufficient accuracy, or when the flight is airborne, i.e. always with enough time

to guarantee a previous Collaborative Decision Making process (i.e. Collaborative Flight

Planning). It excludes tactical instructions and clearances that need an immediate response,

but includes activities such as dynamic route allocation [1].

Optimal profile: constant and progressive elevation and descent (no steps).

Traditional profile: staggered elevation and descent (with acceleration steps).

Figure 4: Comparison between the optimal vertical profile and the traditional one



- 10 -

In this context, strategic conflict management and traffic synchronization, would lead to pre-

deconflicted 3D routes subject to dynamic refinement or adjustment during flight (i.e. 4D

contracts), as a result of the introduction of automation support to conflict detection, situation

monitoring and conflict resolution, the main initiatives for increasing airspace capacity in the

period up to 2020. Thus, the SESAR’s role to achieve these capacity perspectives (extra

capacity is required since an increment of the current air traffic levels in a factor of x2 is

forecasted by 2030) is to deploy tools to assist the controller with complex situations and to

reduce complexity by strategic deconfliction measures.

A progressive improvement in the accuracy of ground-based trajectory prediction through

reduced flight uncertainty will lead to improved performance of controller support tools. The

consideration of 4D trajectories together with the representation of wake vortex with ECAC-

Wide perspective is considered necessary in this work to allow 4D trajectories, to be planned

safely with regards the potential hazards of WVE during all phases of flight.

State of the art - Wake Vortex Phenomenon

- 11 -

2.2 Wake Vortex Phenomenon

2.2.1 Origin and stochastic behaviour

All aircraft produce wake turbulence, also known wingtip vortices or wake vortices

when they are airborne during a flight. Wake vortices are formed any time an aerofoil is

producing lift. Lift is an upward force generated by the creation of a pressure

differential over the wing surfaces. The lowest pressure occurs over the upper surface of

the wing, and the highest pressure is formed under the wing. Air will always move

towards the area of lower pressure. This causes it to move outwards under the wing

towards the wingtip and curl up and over the upper surface of the wing. This starts the

wake vortex (see Fig. 5) [15, 17].

The same pressure differential also causes air to move inwards over the wing. Small

trailing edge vortices, formed by outward and inward moving streams of air meeting at

the trailing edge, move outwards to the wingtip and join the large wingtip vortex.

Swirling air masses trail downstream of the wingtips. Viewed from behind, the left

vortex rotates clockwise and the right vortex rotates counter-clockwise (see Fig. 5).

Typically, a vortex develops a circular motion around a core region. The core size can

vary in size from only a few centimetres in diameter to a metre or more, depending on

the type of aircraft. From larger aircraft, the speed of the air inside this core can be up to

100 metres per second. The core is surrounded by an outer region of the vortex, as large

as 30 metres in diameter, with air moving at speeds that decrease as the distance from

the core increases. Wake vortices can persist in the air for three minutes, or longer, in

certain conditions [15, 17].

Figure 5: Wake vortex roll-up process

Source: [15]


- 12 -

In order to obtain a deeper knowledge about the nature of this phenomenon for the

purpose of this work, it is important to analyse its characteristics and its development

and evolution over time. These characteristics are:

Circulation

The strength of the vortex generated after this roll-up process, as a result of the

combination of same direction vortices, is expressed by the circulation or rotation

velocity.

Equation (1) is the KuttaJoukowsky relation, which is used to determine the initial

circulation, Г0, of a single vortex, with m2/s as unit. If we observe the previous

expression, the circulation depends only on air density (ρ) and aircraft parameters as

weight (W), velocity (VL) and wingspan (bL). The vortices are initially positioned at a

distance of about b0 = bL 𝜋

4 from each other, which is less than the aircraft’s wingspan

[4].

Figure 6: Plan view and lateral view of wake vortex Source: [17]

(1)


- 13 -

Descent

Another important characteristic of the vortices is their tendency to descent from their

initial altitude, phenomenon influenced by buoyancy effects of the atmosphere that

causes a constant upward pushing force. This effect is called rebound. So, we can

conclude that with a stably stratified atmosphere, the buoyancy force additionally

reduces the descent velocity. Equation (2) describes the initial descent (ω0) of a vortex

pair.

It is important to mention that buoyancy force depends on the stratification of the

atmosphere and, at the same time, this last one, is dependent of the temperature

gradient. Thus, a positive, or small negative temperature gradient will induce a large

buoyancy force, making the wake rebound, while a large negative temperature gradient

will decelerate the wake’s descent. The typical pattern of wake vortex in standard

atmosphere is shown in the Fig. 7 [4].

Decay

After this so-called roll-up process, the main trend is that the vortex strength is reduced

by the time. This process depends on the internal characteristics of the vortex and

external characteristics of the surrounding atmosphere. This phenomenon is produced

when the stably rotating airflow is broken down, as a result of the interaction with

internal and external vorticities. The decay of the vortices occurs in two phases, as is

shown in Fig. 8.

Figure 7: Typical pattern of wake vortex in standard atmosphere

Source: DE CLEYN, 2011, [4]

(2)


- 14 -

In the first phase, called turbulent diffusion phase, the circulation decays rather slowly.

However, after a certain time T*, which mainly depends on the atmospheric conditions,

the rapid decay phase starts and it is characterized by a higher rate of the vortex decay.

Transport

Considering an atmosphere with non-zero wind, the wakes will move with the wind

after their generation. This is a phenomenon called wake vortex transport. The position

with respect to each other does not vary during the transport, maintaining both in

parallel. The transport of the vortex pair depends on the combination of the velocity

field of the vortex particles and those of the atmospheric velocity components. These

include wind, but also wind shear and stratification [4].

Figure 8: Two phase decay



- 15 -

2.2.2 Potential hazards and safety distances

The International Civil Aviation Organization (ICAO) is an organization that develops

regulations and recommendations for international air navigation. In ICAO’s 6108873-

DOC-4444 documentation different safety distances are described to vertical, horizontal

and temporal level that should be preserved between aircraft.

Horizontal safety distances include the lateral and longitudinal separations. A way to

simplify the ICAO regulation is to assume that all aircraft will be controlled by radar, so

you can take a sufficiently conservative separation measure, which is considered to be

5NM (9,3 km) as horizontal safety distance between aircraft for en-route airspace,

which can be reduced to 3NM (5.5 km) in the case of TMAs [5].

For vertical safety distances a minimum separation of 1000ft (300 m) is currently

defined for those airspaces where a vertical separation called RVSM (Reduced Vertical

Separation Minimum) is applied. This is the case in Europe, where the rules of RVSM

for all levels of civil aviation flight were implemented since 2002, with the idea of

increasing airspace capacity while maintaining the same level of safety [5].

The RVSM procedure was extensively investigated on potential safety risks before

implementation. The probability of an aircraft wake descending 1000 ft below its

generation altitude, while being of sufficient strength to cause major turbulence was

concluded to be very low, but this investigation did not consider a thorough analysis of

real atmospheric circumstances, and the impact of new weight classes such as the very heavy

aircraft.

By modelling the wake vortex transport and decay taking into account a meteorological

forecast, it could be demonstrated that at cruise altitudes, heavy aircraft may produce a

wake that can sink as low as 1000 ft below its generation altitude and remain

sufficiently strong to cause severe turbulence. This statement was verified by a number

of wake vortex encounter incident reports, where aircraft encountered severe turbulence

caused by the wake of a heavy aircraft passing overhead, one flight level above. Among

the meteorological conditions that can favour this phenomenon are a negative

temperature gradient, low atmospheric turbulence and wind velocity.

Therefore, it is very important to revise these safety distances, because wake vortex

encounters can cause serious consequences. Reported wake vortex encounter incidents

indicate that trailing aircraft may suffer very short period vertical accelerations between

-1 and 3g. These often result in altitude excursions and violent roll accelerations, which

may bring an aircraft in conflict with other traffic on higher or lower flight levels. The

combination of vertical, lateral and roll accelerations can bring crew or passengers that

are not strapped down by their seatbelt out of balance and can cause severe injuries. If

these rolling motions occur in close proximity of the ground, there may be no more time

to recover and avoid impact with the ground.


- 16 -

The hazards of WVE (Wake Vortex Encounter) in en-route phase of flight can become

much severe in the context of the modernization of the ATM (SESAR 2020+), in which

a deep revision of the procedures and separation standards is going on. Also much more

demand and less homogenous aircraft mix is expected (i.e. heavier aircraft sharing

airspace with lighter ones), thus increasing the probability of encounter and the severity,

parameter that depends on the rotation velocity of vortex, the stability factors of the

affected aircraft, and the space geometry where these ones occur, being very important

for the purpose of this project [22].

For instance, see Fig. 2 in which airspace and flight operations could be not constrained

by the current Flight Level Scheme that is used to separate traffic (which is in line with

future SESAR ATM).

For this reason, and considering the ATM’s future capacity and safety requirements, we

need new technology and systems capable to know, in every moment, the wake vortex

position and its strength, because it can be very variable depending on the aircraft’s

characteristics and specially on the meteorological conditions.

Figure 9: Wake Vortex Encounter


State of the art - SDS and their use in to tools for ATM

- 17 -

2.3 Spatial Data Structures (SDS) and their use in to tools for

ATM

In order to develop the Wake Vortex Encounter Predictor System proposed in this

project, we will use the current conflict detection technology based on Spatial Data

Structures (SDS).

A SDS is a tool capable to represent a particular spatial region (in this case an air sector)

being a very easy way in terms of computational treatment. Actually, a SDS is a

database in which every memory position represents each of the discrete coordinates of

the modelled sector. From a conceptual point of view it can be thought as a mesh of

discrete points distributed throughout the space region, in which the spatial information

are stored (associated with such a coordinate) making easy the access to this data,

applying simple mathematical formulas.

Fig.10 illustrates this concept. Note that inside this three-dimensional SDS, represented

by a cube, a discretized 4D trajectory is stored (different 3D positions of an aircraft in

different discrete time steps).

The tube that envelops the trajectory establishes the space that should be reserved for

each trajectory in the SDS, considering the safety distances between aircraft and the

behavior of wake vortex that can cause instability if there is an encounter. Given that a

wake vortex is usually propagated down 1000ft (300m), the radius of the tube is 300

meters.

To understand properly the system, it is necessary to know the granularity concept. The

granularity or resolution of a SDS is the distance between the constituent discrete

points of the SDS.

Figure 10: SDS conceptual representation

Source: Ruiz, Piera, 2013, [1]


- 18 -

Depending on the granularity chosen we will have a specific quality referring to the

state-space stored and a certain degree of efficiency at processing and managing the

spatial data. Therefore, an excess of resolution may lead to a loss of computer

performance as well as to an inoperable amount of memory requirements, whereas a

lack of resolution may lead to lose some important objects of the space [1].

So, to determine the appropriate resolution, taking into account the region that we want

to model, we have to consider the next factors among others:

The size of the physical airspace to model

The size of the objects to be stored in the database

The speed at what these objects move

The quantity of memory available in the computer

The speed of execution of the algorithms

The SDS should be implemented as a large one-dimensional array stored in a RAM

memory, which can be conceptually drawn as a table that contains as many rows as

coordinates are in the modelled airspace and as many columns as aircraft / processing

trajectories are. See Fig.11 to observe an example of spatial data and a specific SDS

configuration.

The data stored in the SDS consists of both spatial information, like the discrete

representation of either the trajectory or its corresponding enveloping safety tube, and

non-spatial information like the identification number of the flight and the timestamps

associated with each discrete trajectory or tube, which can be stored explicitly in one or

several fields in the same database record.

The representation of the discrete airspace coordinates by each row/record and their

sequential classification in the database, make easy the access for writing/reading

applying the following equation:

SDSrow (x, y, z) = x · Y + y · Z + z + 1 (3)

being SDSrow the record position inside the database that stores the information of a

particular coordinate (x, y, z), with x ∈ [0 , X − 1], y ∈ [0 , Y − 1] and z ∈ [0 , Z − 1],

and being X, Y and Z the maximum number of discrete coordinates of each spatial

dimension, respectively.


- 19 -

Note that in Fig. 11, the information stored in each record is about the aircraft/flight that

will use the corresponding coordinate and the time period in which this one will be

occupied. In the fourth record it can be found that aircraft with id = 2 is expected to

make use of the coordinate (0, 0, 3) in the time window from second 520 until second

640.

As you can see, the coordinate (0, 1, 1) is used by two different aircraft in overlapped

time periods (TWon refers to the moment when the aircraft occupies the coordinate and

TWoff to the time in which the aircraft leaves) information that allows the detection of a

conflict between both in this time-space. This is the mechanism used by current systems

based on SDS to detect conflicts.

However, there are coordinates that are not booked for any aircraft, and this system

requires as much space for these coordinates as N reservations are done, causing an

unnecessary increase in memory use (Examples: (0,0,0) and (0,0,1)).

The use of SDSs allows the storage of the entire State-Space description of the air

traffic at a given time and its evolution over the time. All the processed trajectories will

remain stored as a “4D snapshot” of the ATM system which makes the SDSs to be an

interesting tool for traffic separation purposes since:

• The SDS acts like a powerful spatial pruning filter which it makes available a

continuous updated air traffic state due to its capability to process a large number of

trajectories within a large time-window look-ahead (e.g. 2 hours) in few seconds, which

allows to tackle some ATM sources of uncertainty.

Figure 11: SDS content example



- 20 -

• The SDS can be used through causal models in order to efficiently access and explore

the SS information stored inside, which enables an improvement of the decision-making

with a global optimization scope due to the capacity to evaluate all the potential network

emergent dynamics.

2.3.1 SDS applications for ATM purposes

Spatial Data Structures have been explored in different previous researches as a highly

configurable technique easily used in different ATM applications, whose

implementation to store the relevant state-space presents excellent results in collision

avoidance algorithms, in time-based MTCD prototype for Terminal Maneuvering Areas

and in the strategic conflict detection and resolution process of thousands of trajectories

at European ATM level.

2.3.1.1 Collision Avoidance

A common method to detect conflicts between trajectories is the application of pairwise

CD algorithms, which consist on distance calculations between the point-mass positions

occupied by the aircraft at each given time-step, comparing each different pair of 4D

trajectories. In Fig.12 we can observe how the dotted lines are representing the 4D

coordinates that belong to the same time instant.

However, these pairwise algorithms are not efficient because of its computational

complexity of quadratic order, o (n2), thus it is necessary the use of some kind of filters

(e.g. flight level pre-filter, time-skipping strategies. . .) in order to avoid unnecessary

comparisons, improving the time performance of the algorithms.

Figure 12: Representation of a pairwise algorithm



- 21 -

As an alternative to this method, we can use SDSs as a powerful spatial pruning filter

that linearizes the runtime/complexity, o (n), of the collision avoidance algorithm. This

one, considers the fact that collisions can only occur when two boids are at a certain

short distance, i.e., when agents are geographically correlated.

Thus, the characters “pre-sorted” are kept in the SDS depending on their location in

space, allowing a quick identification of the neighbourhood in which they are at a given

time-step, achieving a reduction of the amount of these pairwise computations between

trajectories. Therefore, firstly, the neighbours are identified at each time step (i.e. boids

geographically correlated at close locations) in order to filter the amount of pairs to be

compared with and, after that, the spatial distances among those still-remaining pairs of

trajectories are checked, performing the detection of conflicts for each boid [1].

Fig. 13 shows this concept. A boid is circled representing its safety area. Its associated

neighbourhood are formed by the grid-cells overlapped with the circle (shadowed) at the

time of simulation. If there are boids inside the neighbourhood at the same time instant

of simulation they will be candidates for a pairwise distance comparison. Note that Fig.

13 represents a top-view of a 2D scenario, but the same concept can be extended to 3D.

This algorithm has the capability to update the content of the SDS at each simulation

time-step so that the temporal information is not stored. Therefore, the algorithm

considers a SDS configured to store only the id information of the boids that are located

over the surface of the grid cells (the bin-volume in case of a 3D scenario), determining

the next trajectory-position. Consequently, if at current time of simulation a

conflict/potential collision is detected, the boids react by changing their trajectories at

next time of simulation.

Figure 13: Neighbours search to filter some of the pairwise comparisons



- 22 -

However, when a large amount of boids is contemplated, the emerging dynamics effects

can make the avoidance of collisions unfeasible. Therefore, the whole algorithm and

SDS architecture must be reconfigured for ATM planning applications, such as strategic

de-confliction and/or tactical management of conflicts in the European airspace, in

which a look-ahead planning horizon is required and domino effects must be

considered, together with ATM uncertainties.

2.3.1.2 MTCD based on SDS

The technique explained above based on SDS has been applied to obtain a MTCD

algorithm for Terminal Maneuvering Area (TMA). This one has the function of check

time-distance separations between approaching/landing aircraft according to the wake

vortex generated by other aircraft. So, the SDS is used in this algorithm in order to take

a 4D snapshot of the scenario to observe the trajectories executed by aircrafts that are

flying over the modelled area and the vortices generated.

In this particular CD algorithm, the SDS’ logical architecture allows the booking of

spatial discrete cells by each aircraft/flight according to their expected vortex

dimension, concept shown in Fig.11 as the storage of the reservations of resources at

each database record.

The SDS configuration used in this algorithm is different that the used in the conflict

avoidance algorithm mentioned previously. In this last one each SDS record represents

a grid area/volume, also called bin, while MTCD algorithm treats every record as a

single discrete coordinate, i.e. a spatial resource that only can be used by one aircraft at

a given time, considering a fixed time window of 120 seconds for each flight wake

vortex’s duration to simplify.

The process followed when we are using the SDS to detect conflicts, is the next one [1]:

- Firstly, at the moment of storing the surface tube-point, we have to identify its

corresponding coordinates, and then, the algorithm checks if the coordinates are

already occupied. Thus the first column is read and if its value is equal to zero, i.e. is

empty, it means that no other aircraft intend to use such a coordinate, allowing the

storage of this spatial resource as a conflict free position (e.g. in Fig. 11 aircraft #15

could freely reserve (0, 0, 2) since there was no previous reservation).


- 23 -

- If the first column is not empty, the coordinate will be used by another aircraft in a

certain period of time. For this reason, the algorithm compares the time windows to

check if they overlap, in which case a conflict is detected and the Conflict

Resolution (CR) system is informed (e.g. in coordinate (0, 1, 1) there is a conflict

between aircraft #3 and #4 because their utilization time-windows overlap). If the

time windows are not in conflict, it means that the coordinate might be booked in

the following column without informing the CR (e.g. in (0, 2, 0) aircraft #15 could

make a reservation with no conflict with previous reservation of aircraft #6).

In next columns the same procedure is applied sequentially (e.g. firstly, the coordinate

(1, 0, 1) was booked by aircraft #4 and later by aircraft #15. They were not in conflict

because their utilization time windows did not overlap. However, when a third aircraft

#19 made a reservation for the same coordinate, the algorithm detected a conflict with

aircraft #4, without affecting the second one #15, which was still conflict-free).

The implementation of SDSs as efficient spatial information managers, capable to

provide a 4D snapshot to analyze the scenario evolution, makes it an interest tool for

ATM planning purposes. This is a conclusion obtained through several simulations in

Gran Canaria TMA, validated with B738. However, the initial logical architecture of

SDS (shown in Fig.11) is not prepared to consider wide airspaces and/or large amount

of trajectories, due to its lack of efficiency in the computer main memory management,

which makes it impractical for strategic European ATM dynamic planning.

2.3.1.3 Strategic Deconfliction based on SDS

SESAR WP-E launched a project called STREAM (Strategic Trajectory de-confliction to

Enable seamless Aircraft conflict Management) whose objective is the development of

innovative computational-efficient Conflict Detection and Resolution (CD&R) algorithms in

order to apply strategic de-confliction considering thousands of trajectories and taking into

account Airspace Users (AUs) preferences and network constraints, within just few seconds

or minutes. This is aimed at enabling traffic to be de-conflicted for wide airspace regions and

permitting large look-ahead times of order of hours (e.g. two or three hours) [1].

Strategic de-confliction STREAM algorithms make possible the building of a micro-model

framework that represents all the traffic at European airspace scale which allows its efficient

management through a set of individual 4D business trajectories and by suggesting

strategically de-conflicted trajectories which closely match AUs preferred ones in a free-

route environment, i.e. not constrained by pre-structured routes as occurring today.

Under the STREAM project, Spatial Data Structures were explored as a tool with capability

to efficiently store spatial data (e.g. 4D trajectory information) at the time when the conflict

detection among all SBTs/RBTs is performed, allowing an effective Conflict Detection (CD)

process, with excellent results in terms of time performance.


- 24 -

A part from that, the efficient database access methods, have been a key factor for the

development of new tools to analyze the entire ATM State-Space information under a global

scope.

This enables for example a complete and precise identification of the domino effects that the

emergent dynamics may cause in the network, because all the processed trajectories will

remain stored as a “4D snapshot” of the European ATM [1].


- 25 -

2.3.2 SDS enhancements to reduce the memory and time

In spite of the benefits of SDS-based CD algorithm in terms of information

management, this one presents some shortages in its scalability to manage the whole

European ATM. For this reason, new SDS concepts were created considering two

innovations explained below [1]:

• Relational SDS (RSDS): allows a reduction of approximately 98% in terms of main

memory required by original SDS architecture, being a key factor to enable the storage

of the overall European ATM 4D state-space with a look-ahead of several hours. In

addition, the computational time performance and the rest of the advantages of the

original algorithm are maintained.

• Time-Space Data Structure (TSDS): adding a fourth dimension (i.e. the time) to the

original SDS architecture, an efficient management of 4D data and an improvement of

the run-time performance of the CD process is achieved when a large amount of

trajectories is considered. In order to efficiently adapt the ATM planning to

uncertainties, perturbations and system disruptions, the fast processing time achieved

referring to the CD process becomes to a key factor, since the whole state-space can be

updated in real-time.

2.3.2.1 Relational SDS (RSDS)

The SDS-based CD algorithm presents an immense growth rate of memory required

when the number of trajectories to be processed is increased, constituting an important

shortage that drastically restricts the size of the modelled sector and/or the granularity of

the SDS and/or the amount of trajectories that can be analyzed. Therefore, it seems

necessary the use of the memory in a more efficient way.

In order to calculate the necessary amount of memory to store the logical SDS

architecture, we have to know the number of constituent database records (i.e. rows),

which is illustrated in Fig.11. The total number of rows is the result of multiply the

number of records corresponding to each spatial dimension, X · Y · Z, which is

obtained dividing its length (i.e. lengthX, lengthY and lengthZ) by the size of each grid

cell or bin, i.e. SDS granularity [1]. A part from that, the SDS’s configuration allows the

storage of N potential reservations for each record, which occupy B bytes each in the

main memory (B= 4 bytes in the MTCD application above). Thus, the total amount of

memory required to allocate this SDS architecture is:

(4)


- 26 -

Since a new trajectory is introduced into the problem at each time instant, the SDS’

memory positions (of size B each) are increasing continuously in an amount equal to the

number of rows required to represent the airspace scenario of interest, which is usually

huge due to the airspace sizes needed for ATM applications.

Moreover, note that when we are working with three-dimensional granularity, the

amount of memory required grows cubically (e.g. reducing the size of the bins in a

factor of 10 in each spatial dimension implies a memory growth of order 103), which

makes impractical the allocation of the SDS in the current computers’ main memory.

In addition, the SDS architecture presents another important shortage which is the

impossibility that an aircraft can use a same coordinate more than once (e.g. during

holding trajectories), due to the logical architecture’s configuration that only allows a

reservation per aircraft at each SDS record.

A part from that, there is another source of inefficiencies: the amount of empty SDS’

memory positions, i.e. that ones that are not storing any king of useful information. The

result is a SDS with a lot of fields setting to zero that are consuming memory

needlessly, since the amount of memory required for a potential reservation of any

coordinate/record has been already allocated.

A way to achieve more efficient information management is by using different

databases, one optimized to store the basic SDS’ structure, i.e. the memory positions

that configure the modelled airspace to manage the spatial data efficiently, and another

one optimized to store the associated non-spatial information of the

trajectories/reservations. Using this method we are minimizing the immense growth of

the SDS due to the fact that the information of those databases can relate each other

through database pointers to outside database records, becoming into relational

databases (see Fig.14) [1].

To achieve the equivalent functionality of non-relational SDS (Fig.11) these two

databases are required:

Figure 14: Relational SDS architecture



- 27 -

The Base SDS (BS): is a database formed by an amount of records equal to the

number of discrete coordinates of the airspace, i.e. with the same amount of records

than in Fig. 11 (i.e. one row for each discrete coordinate). However, in this case the

BS uses only 1 field/column per record (usually occupying 4 bytes) instead of N

fields/columns for N trajectories (4 · N bytes). The content of this unique column

may be zero if the coordinate will not be occupied by any aircraft or may store a

pointer to a record position of the second database in the contrary case.

The Stacked Trajectory Information (STI): the advantage of this database is that

only stores that coordinates going to be used by all the trajectories, eliminating one

of the most important sources of inefficiencies. However, the most important aspect

to mention about this one, is its particularity to store all the information about

trajectories in a stack (i.e. FIFO order), allowing the optimization of the information

storage since no empty records/memory-positions are present at this database

(saving lot of memory with respect the non-relational SDS). Note that in Fig. 14 the

STI is constituted for 4 columns. In this example, 3 of them are used to store the

non-spatial information of a booking, as it happens in the non-relational one.

However, the column number 4 stores a pointer to another STI position, if later

reservations for the same coordinate are made by same or different trajectories.

Therefore, is a method to establish the order of the coordinates’ use and a way to

know how many aircrafts will use each one of them.

If you compare the information included in both SDSs (Fig. 11 and 14), you will note

that they store the same spatial and non-spatial information. If the pointer to STI stored

in BS is not zero (see Fig.14), at least one booking was done for this coordinate. The

value of the pointer indicates the position of the previous booking stored in STI (e.g. for

coordinate (0, 1, 1) a pointer is stored to position 5 of STI). In order to know if there is a

conflict between the current and the previous booking you only have to access to this

record/row in the STI and check if their utilization time-windows overlap.

After the checking performed to know if there is a conflict with such a previous

booking, the next step is to search if another pointer to STI is present in the fourth

column. If the pointer is set to zero, it means that it does not exist another previous

booking for such a coordinate, allowing the storage of the current booking in the last

free position of the STI, always following the FIFO order. Moreover, the pointer to that

record (the new booking) is stored in the fourth column of the actual current record of

the STI. However, if the pointer is not zero, the algorithm is executed sequentially as

many times as it is necessary until a free position is found to complete the booking (e.g.

in Fig. 14 positions 10 and 14 are sequentially checked) [1].


- 28 -

The main advantage of the RSDS design is its capability to adjust the amount of

memory required used to configure the BS to the size of the airspace sector to model

and the desired granularity, since this architecture allows unlinking its size to the

amount of trajectories considered in the problem. Therefore, with a larger amount of

trajectories, its size does not increase. Thus, the total memory required for the BS

construction is:

being P the amount of bytes required to store a pointer to a record (typically P = 4 or P

= 8) and the rest of the parameters with the same meaning as in Eq. (4). On the other

hand, the amount of memory needed for the STI is calculated by:

(6)

being N the number of trajectories to be processed, L the average amount of time-steps

per trajectory and B and P with the same meaning as in Eq. (4) and Eq. (5). Therefore,

the total memory space needed to store the STI when N increase grows with a constant

linear rate much lower than in the case of the original logical SDS architecture.

The results obtained through several RSDS configurations adapted to different ATM

scenarios, have demonstrated the memory management improvement in terms of

important reductions in the quantity of necessary memory. This reduction are comprised

between 95-99%, as the most ATM practical cases show, which implies a much more

efficient way to manage the main memory in comparison with the use of non-relational

SDS for the same scenario.

The 4D snapshot of the ATM is still available in the RSDS showing all the state space

of the problem, being possible to extract and summarize useful information to use the

new CR algorithms in the necessary cases in order to find efficient and optimal conflict-

free trajectories.

2.3.2.2 Time-Space Data Structure (TSDS)

The kind of SDS for collision avoidance introduced in section 2.3.1.1 had the capability

of storing the trajectories’ information depending on the grid cell/bin that they were

occupying at a given time instant of simulation. This information was useful to detect

how many pairwise comparisons were necessary to perform, acting as a pruning filter

that improved the performance and the efficiency of the CD process.

(5)


- 29 -

However, for ATM planning purposes a similar CD algorithm could be contemplated,

as an algorithm capable to refresh the SDS at each time of simulation and to create a

registry composed by the detected conflicts that the CR module will have to solve.

Nevertheless, in order to have a network global view of the user’s flight intentions and

to have more capability to predict the potential emergent dynamics, a historical record

of the evolution of all trajectories over the time (like a 4D snapshot) is highly beneficial

and necessary to guarantee a more efficient management of the European air traffic.

To obtain the 4D ATM snapshot, we can modify the SDS’ configuration to store the

expected time-window in which each coordinate will be used at each cell/bin

reservation, in a similar way than seen in Fig.11. However, this option has a shortage

since the benefits of using such SDS configuration decrease when there is a large

amount of trajectories willing to use the same SDS cells/bins, even in different time

instants. This shortage proceeds from the computational cost that the comparisons with

previous trajectories suppose, even if they are filtered by the time-window of utilization.

By adding the temporal dimension (the 4th one) to the SDS’ structure, we can achieve a

cost reduction of comparing with previous trajectories, since that reservations for

different expected times of utilization, even for the same spatial resource, are considered

as different reservations for different time-space regions. Therefore, this method

becomes to a more powerful filter than only using spatial queries, since by the

utilization of time-spatial ones we can achieve a higher reduction of the pairwise

comparisons [1].

As a result, a new SDS configuration is obtained named Time-Space Data Structure

(TSDS). Conceptually, a TSDS can be thought as a set of T different SDSs, one for each

discrete portion of time (see Fig. 15). The TSDS configuration has the capability to

support time-windows as discrete portions of time, not only considering time-instants

(Fig. 15 shows this concept, since a set of T different SDSs, each one storing 4 time-

steps of different 4D trajectories executed in different time-windows, is illustrated).

Figure 15: TSDS conceptual representation



- 30 -

The granularity of the temporal dimension (i.e. lengthT sizeBinsT), will depend on both

the order of such dimension (i.e. length of t axis) and the amount of discrete portions of

time (i.e. size of cells/bins in the temporal dimension), requiring the following amount

of memory to store the TSDS:

The meaning of parameters in Eq. (7) is the same as in Eq. (4). The combination of the

concepts of the RSDS and TSDS is also possible (RTSDS), thus reducing the memory

needs to support the TSDS (similar to Eq. (5)):

In the logical structure of the TSDS each 4D coordinate is represented by a single

database record/row, sorted sequentially to ease the reading/writing content access. The

access method to records of a (3D) SDS is extended to take into account the 4th

dimension:

Thus, inside the TSDS, the SDSpos represents an univocal record/row position that

stores the information relative to a given 4D certain coordinate (x,y,z,t), while X, Y, Z

and T are the total amount of different discrete values that the constituent variables of

such a 4D coordinate can adopt, according to the order/size of each respective

dimension.

Figure 16: Neighbourhood (shadowed) defined by geometrical arguments

total Memory SDS = X·Y·Z·T·B·N = 𝐥𝐞𝐧𝐠𝐭𝐡𝐗

𝐬𝐢𝐳𝐞𝐁𝐢𝐧𝐬𝐗 ·

𝐥𝐞𝐧𝐠𝐭𝐡𝐘

𝐬𝐢𝐳𝐞𝐁𝐢𝐧𝐬𝐘 ·

𝐥𝐞𝐧𝐠𝐭𝐡𝐙

𝐬𝐢𝐳𝐞𝐁𝐢𝐧𝐬𝐙 ·

𝐥𝐞𝐧𝐠𝐭𝐡𝐓

𝐬𝐢𝐳𝐞𝐁𝐢𝐧𝐬𝐓 ·B·N (7)

total Memory BS = X·Y·Z·T·P = 𝐥𝐞𝐧𝐠𝐭𝐡𝐗

𝐬𝐢𝐳𝐞𝐁𝐢𝐧𝐬𝐗 ·

𝐥𝐞𝐧𝐠𝐭𝐡𝐘

𝐬𝐢𝐳𝐞𝐁𝐢𝐧𝐬𝐘 ·

𝐥𝐞𝐧𝐠𝐭𝐡𝐙

𝐬𝐢𝐳𝐞𝐁𝐢𝐧𝐬𝐙 ·

𝐥𝐞𝐧𝐠𝐭𝐡𝐓

𝐬𝐢𝐳𝐞𝐁𝐢𝐧𝐬𝐓 ·P (8)

SDSpos = x · Y · Z · T + y · Z · T + z · T + t (9)



- 31 -

2.3.2.3 RTSDS: Combination of RSDS and TSDS

The combination of the two SDS configuration introduced above, i.e. RTSDS

configuration (RSDS + TSDS), was tested in the STREAM project [1] in order to build

a strategic Conflict Detection & Resolution tool. The tests have only taken in

consideration the route phase and the granularity used is 20 km x 20 km x 600 m, which

is in reality the dimension of the constituent bins. The utilization of these dimensions is

because these ones are the double of the minimum safety en-route separation defined in

the current ATM, i.e. 5NM (∼9.3 Km) in the horizontal plane and 1000ft (∼300 m) in

the vertical one. Since the second is the time unit used to discretize the 4D trajectories

of the aircraft, the resolution of the temporal dimension has also been set to 1 second,

facilitating the construction and manipulation of the RTSDS [1].

The size of these bins has been established considering a trade-off between the resulting

benefits of the RTSDS trajectory pruning (too big bins means a less powerful

“trajectory pruning”) and the amount of RTSDS accesses required for neighbourhood

queries which affects directly to the time consuming (too small bins require searching

for neighbours in more bins).

Taking into account the bins dimension in the horizontal plane (20 km x 20 km), we

have to consider only 4 bins to be accessed in order to complete the neighbour search at

each trajectory time-step (as Fig. 16 shows), while a considerable amount of pairwise

comparisons is filtered due to the relatively narrow dimensions of the neighbourhood

(40 km x 40 km). In the vertical plane occurs something similar since 2 bins are

necessary to ensure the detection of any conflict due to the 600 meters fixed as the bin-

size in this dimension. In total, the algorithm checks 4 × 2 = 8 adjacent bins, looking for

neighbours at each time-step of a given trajectory.

When another aircraft is found inside the neighbourhood formed by these 8 bins (in the

same time instant, since TSDS neighbourhoods are time-spatial regions), in order to

detect if there is a conflict between them, the algorithm performs a direct distance

comparison between their expected point-mass positions. In this case we will use an

airspace sector of 5.000 x 5.000 km2

of surface, i.e. the most part of the European ATM,

as an example to show the amount of memory required using this kind of SDS

configuration (RTSDS). Therefore, 20 flight levels (6000 meters) and a strategic look

ahead of 5 hours with a resolution of 1 second will be considered in the CD process.

Then, the memory size occupied by the BS is (consider 4 bytes per row) [1]:

X = Y = 5000km/20km = 250 (10)

Z = 6000m/600m = 10 (11)

T = 5h · 3600s/h = 18000 (12)

BS= X·Y·Z·T·4 = 250 · 250 · 10 · 18000 · 4 = 45GB (13)


- 32 -

Let consider a maximum of 30.000 different trajectories within the 5 hours look-ahead

of the scenario, with average flight duration of 2 hours (and resolution of 1 second).

Then, the STI will occupy (consider 8 bytes per row)

As you can observe, the total memory required using the RTSDS and considering this

relatively wide scenario described before would occupy less than 47GB, an amount

easily supportable by the current commercial 64-bit computer in RAM main memory

(instead of using external hard drivers that are much slower at reading and writing the

information). A CR module can take advantage to the capability of the RTSDS

configuration since it makes possible the storage of the ATM 4D state-space within 5

hours strategic look ahead. It means that the content of the RTSDS can be updated in

order of seconds, a strategic key factor that can be used to dynamically adapt the flight

routes allocation in response to network changes, which allows a balance in real time of

the actual available airspace capacity and demand.

STI = N · L · 8 = 30000 · 2 · 3600 · 8 = 1.8GB (14)

Wake Vortex Encounter Prediction System

- 33 -

3. WAKE VORTEX ENCOUNTER PREDICTION

SYSTEM

The SDS architectures introduced in the state of art have only been implemented to

enable wake vortex encounter predictions in reduced regions like TMAs, but the future

ATM necessities require the development of a system capable to consider the traffic of

all European airspace. A part from that, it is necessary to introduce information about

the origin and behaviour of the wake vortex in time-space to achieve an efficient wake

vortex encounter prediction system. These two aspects represent the development of

new models and the use of a larger amount of memory, which is currently bounded by

technological levels achieved in computer industry.

Therefore, taking into account the improvements achieved by using Relational SDSs, it

is logical to consider this system as a starting point.

Analysing the RSDS to find a way to include more information without increasing the

necessary amount of memory, we observed the existence of a large set of regions in BS

(memory positions) that were actually empty (i.e. 0 reservations like in coordinates (0,

0, 0) and (0, 0, 1)) during the representation of traffic at ECAC level (more than 4000

trajectories).


- 34 -

Therefore, the main goal of this section is to propose a method in order to allocate in

memory only the regions in which there is a considerable probability that a wake vortex

encounter occurs, and only declare these ones in BS. This concept is the base of the

system proposed in this project to perform Wake Vortex Encounter Prediction (WVEP)

at ECAC-level and with thousands of trajectories. The proposed system architecture is

shown in Fig. 17. SDS will be used in two different processes (HRRI and WVE Risk

Analysis) with different configurations to achieve the goals of this project.

Firstly, the system will process the Flight Plan (FPL) of all the flights scheduled to

provide information about the origin, destination and the route preferences for each

flight. These flight plans are used by the Trajectory Predictor (TP), where they will be

processed in order to obtain the associated 4D trajectories with their corresponding

waypoints. This TP module is assumed to provide optimal trajectories for the airline

operator.

Then, the High Risk Region Identificator (HRRI) will be fed with the outputs of the

Trajectory Predictor to detect the regions or neighbourhoods where there is a high

probability of a severe wake vortex encounter. The identification of these HRRs (High

Risk Regions) will be performed with the help of SDSs.

To avoid excessive and unnecessary computational effort, the wake vortex will be

simulated by the WV Simulator module only in those regions considered dangerous.

Once the wake vortex is simulated for each of the identified regions, it will be processed

by the WVE Risk Analysis module. Here a configuration with SDSs will be also used in

order to efficiently and accurately detect potential wake vortex encounters. If a wake

vortex encounter is predicted then the associated severity will be calculated, to

determine the actual risk of such encounter.

So, to summarize the procedure that the system follows, in the next figure you can

observe the associated flowchart of the proposed system, which includes all functions

explained above and the way in which they are interconnected.

Figure 17: Wake Vortex Encounter Prediction System's Architecture


- 35 -

Yes Yes

No No

Yes

END PROCESS

Expected TRs

WVE Risk Analysis

(Probabilistic

Prediction)

WV Simulator

HRRI Pre-process

WVE

Probability

?

Severity Inform user or

WVER

More

TRs?

No

Figure 18: System's Flowchart

4D Trajectories

Regions + Trajectory segments

Regions

+ Trajectory segments

+ Associated WV/4D tubes

WVP System


- 36 -

The modules that use SDS configurations to perform their functions are those explained

in the next sections (3.1 and 3.2), which are the base of the proposed high-level

architecture.

3.1 High Risk Region Identificator (HRRI)

To perform this function properly and efficiently, it is necessary to analyze the future

ATM situation and to assess if the current safety standards are suitable for the future

necessities. Forecasts indicate an increasing traffic density, whereby the prediction of

encounters with enough time ahead will be essential.

This module is necessary to filter unnecessary computations when the process to

identify the regions of interest is being performed, improving the system’s efficiency

and its time-performance.

The current safety standards are 5NM in the horizontal plane and 1000ft (300m) in the

vertical one, as Fig. 19 shows. However, these distances are sometimes not enough to

protect flights against wake vortex encounters [4]. Therefore, in this project we will use

10 NM and 2000ft as conservative distances which contemplate all the possibilities

referring to the behavior of wake vortex including the uncertainties that can be present,

thus allowing a prediction with enough time ahead that guarantees the effectiveness of

the system in terms of safety.

Figure 19: Current Safety Standards


- 37 -

The process to filter unnecessary computations consists in performing a pre-process

considering the same SDS configuration (i.e., RSDS and TSDS) than the used for

Strategic De-confliction explained in Section 2, whose objective is obtaining a

microscale snapshot of all European air traffic. Thus, the SDS’s granularity will be 20

NM x 20NM x 2000ft (37km x 37kmx 600m) and we will use the conflict detection

function in order to detect the trajectories that are spaced apart less than 10NM

horizontally and 2000ft vertically (the SDS’ granularity is the double of the distance

considered as safety standard).

In Fig. 20 you can see how each cell has four quadrants, so the first step in the pre-

process will be the identification of the quadrant in which the separation between both

trajectories is less than 10NM, i.e. the affected quadrant. The way to obtain this

information is applying the next equations given a coordinate (x’, y’, z’), each one for

each dimension:

20 NM

20 NM

< 10NM

Tr1 Tr2

Figure 20: Pre-process scenario


- 38 -

Substituting the coordinate values in these equations a value between 0 and 1 will be

obtained, which will give information about the quadrant of the bin occupied by a

certain object (e.g., an aircraft). If the value is less than 0.5, the affected quadrant will

be the first that forms the grid on that axis, while if it is equal or greater than 0.5, will be

the next one.

Once the affected quadrants have been detected, we have to identify the 8 neighboring

quadrants in which the wake vortex will be present (4 horizontally and 2 vertically,

forming a cube).

This method will allow the identification of the regions with a significant risk

probability that an encounter occurs, and it is similar than the used in collision

avoidance applications introduced in section 2.3.1.1. Therefore, the output obtained

through this pre-process is the following one:

The output shows the involved trajectories (first and second columns) and the affected

quadrants as well as neighboring ones, which will configure the High Risk Regions

(HRR). This output is stored in the main memory together with the scenario (to analyze

the changes in the network and the corresponding updated information).

1 2

2 5

… …

Affected Quadrant

Affected Quadrant

…

Neighboring Quadrants

Neighboring Quadrants

…

Pre-process Output

(x0, y0, z0) (xf)

(yf)

(X’, Y’, Z’)

Figure 21: Method to determine the affected quadrant

Figure 22: Pre-process Output


- 39 -

Therefore the TSDS configuration is applied to continuously obtain updated

information of the state-space when the pre-process is performed, following the

trajectories evolution at each given time step in order to identify the HRRs. Apart from

that, the RSDS is also used to create the scenario for the pre-process with all the

necessary information about trajectories, and time-windows associated with each one of

the coordinates occupied. Thus, the union of these two configurations (TSDS + RSDS)

enables the effective and efficient identification of the HRRs, which is the first step to

achieve an architecture capable to contemplate all European airspace, focusing the

efforts in the regions of interest to reducing the necessary amount of memory and to

make it computationally feasible.

HRR (High Risk Region)

< 10NM

Tr1 Tr2

Affected Quadrant

20 NM

20 NM

Figure 23: 2D High Risk Region

Figure 24: 3D High Risk Region (cube)


- 40 -

Once, a HRR is detected, the corresponding SDS is created to represent the scenario. As

a result, we will have N High Risk Regions identified, i.e. N SDS with a granularity of

100 meters to accurately reproduce the dynamics of the wake vortices and the severity

of a potential encounter (if this one occurs), a process that will occur in the WVE Risk

Analysis module. Therefore, there will be N BSs and N STIs, whose role will be played

by Wake Vortex Information (WVI), concept explained below.

3.2 WVE Risk Analysis

The information generated by the HRRI module, i.e., the High Risk Regions identified

together with the involved trajectory segments, will be transferred to the WV Simulator.

Its function is to generate the wake vortex in volume, using a discretization of a second

to achieve a representation with optimum resolution for wake vortex prediction and

severity quantification.

Based on the same simplified models used in [1] and explained in section 2.3, the wake

vortices of each flight will be modeled as a 4D tube used as envelope of the worst-case

behavior of the vortex (as Fig. 10 shows) . However, the model will be improved to

represent more precisely the actual dynamics of position and strength.

100 m

100 m

Figure 25: Wake Vortex's Discretization


- 41 -

The new wake vortex model can be represented as a sequence of shorter tubes

enveloping each one of the velocity vectors that configure the wake vortex. Therefore,

for each segment comprised between a pair of coordinates, a tube is built covering

completely the corresponding part of the wake vortex, which will be centered inside the

tube (this one will be as long as the velocity vector formed by the two coordinates is).

Thus, as Fig. 26 shows, a tube can be represented as a succession of circumferences

situated one in front of the other. Therefore, the construction of a discrete tube can be

simplified by forming a single circumference with a certain radius, in order to do as

many copies as we need to achieve the total length of such the velocity vector. Actually,

it constitutes a discrete model of wake vortex, having as a surface, the worst case of a

probabilistic envelop. So the radius considered in the discrete model will be determined

analyzing the outputs obtained doing simulations.

There are several computer algorithms used to create circumferences. One of the most

popular is the Bresenham algorithm, which uses the symmetries and antisymmetries that

occur between 8 octants in which a circumference can be divided. Thus, it is only

necessary the construction of one of the octants in order to shape the rest of octants by

sign changes in its coordinates, obtaining the circumference shown in Fig. 27 [5].

v

Figure 26: WV's Enveloping Tube

Not to scale


- 42 -

Fig. 29 shows the discretization of a circumference in a grid of pixels to allow its

reproduction on a computer screen. A circumference can be represented by a serie of

separate pixels together with a minimum separation between them equal to the chosen

for the discretization of the space model (in this case, 100 meters). Thus, each point on

the circumference can be thought as a position (coordinate) inside the SDS [5].

0 1 2 3 4 5 6 7 8 9 10 x

y

10

9

8

7

6

5

4

3

2

1

0

y = x

Figure 27: Circumference with symmetrical and antisymmetric segments

Figure 28: Octant formed by joint pixels




- 43 -

Once all octants are calculated on the XY plane, a circumference can be obtained as

shown in Fig. 30. And as Fig. 31 shows, a cylinder can be constructed centered on the z

axis by repeating pre-calculated circumferences separated by a distance of 100 meters to

equal or exceed the length of the corresponding speed vector.

This way of constructing a cylinder is very simple to implement in a computer and it

has the great advantage that the circumferences are pre-calculated, so it is totally

efficient from a computational point of view.

0 1 2 3 4 5 6 7 8 9 10 x

y

10

9

8

7

6

5

4

3

2

1

0

y = x

Figure 29: Octant formed by disjoint pixels

X Y

Z


Figure 30: Discrete Circumference



- 44 -

These techniques were already used in [5], however in this project a new model

enhancement is proposed. We want to achieve a discrete model of wake vortex

represented in a single circumference in order to identify at each coordinate, the wake

vortex generator aircraft’s id, the period of time (Δt) ranging from the instant in which

the wake vortex leaves a coordinate (ti) until the instant in which it enters in the next

one (tf) and the circulation of the wake vortex associated to each time-space position,

thus having a more precise wake vortex dynamic model than in [5] but still being much

more computationally efficient than precise fluid dynamic computations, which are not

fully required for wake vortex encounter predictions and hazards analyses. This concept

is shown in Fig. 32.

Γ0

Γ=0 Δt

(x, y, z) id ti tf Γ0

(x1, y1, z1) id ti1 tf1 Γ1

(x2, y2, z2) id ti2 tf2 Γ2

… … … … …

(xi, yi, zi) id tii tfi 0

Wake Vortex Information (WVI)

Z

Y X

Figure 32: Wake Vortex’s Discrete Model


Figure 31: Construction of discrete cylinder


- 45 -

Therefore, each coordinate represents a position of the wake vortex at each time instant,

considering the first one (centered above) as the position with initial circulation Γ0, and

the coordinate located at 180 º respect this one, as the position in which the circulation

(Γ) is equal to zero. Thus, we can know at each position the severity of the wake vortex

to decide if it is dangerous overfly the area or not.

Finally, we have to understand the last function called WVE Risk analysis. This one

works with information from HRRI and with the data originated in WV Simulator, i.e.

the tubes as well as the severity corresponding to each time-space.

Its objectives are basically to determine the risk of flying over a certain area with a

specific wake vortex and detect a potentially affected aircraft and quantify the severity

of a potential encounter. In the last case the information is completely known thanks to

the WV Simulator function, but to determine the affected aircraft it is necessary a

representation of the wake vortex’s evolution at micro-scale as well as the trajectories

involved in the HRR. Thus the system will display the state of the airspace (HRR) in

order to avoid wake vortex encounters that may endanger flight’s safety, to inform the

Wake Vortex Encounter Resolution (WVER) system or the user, and to take more

efficient and safer decisions. An illustration of the information stored in the SDS used at

a given HRR can be represented as in Fig. 33. In this case the RSDS configuration is

applied again to store all the wake vortex information ( stored in WVI) corresponding to

the discrete vortex model.

In this way we achieve the storage and management information at a micro level (WVE

Risk analysis), through a pre-process performed at a macro level (HRRI) thanks to the

suitable combination of these two configurations.

Figure 33: State space of the HRR

Not to scale

Expected Algorithm Performance

- 46 -

4. EXPECTED ALGORITHM PERFORMANCE

In this section it is anticipated the memory performance of the algorithms proposed for

the Wake Vortex Encounter Prediction system at European scale. The time performance

will be analyzed in future work when the proposed algorithms will be implemented.

It is important to point out, for comparison purposes, that trying to store the relevant

micro-level model of wake vortices at European airspace scale with the RTSDS

configuration seen in section 2.3.2.3, will lead to untreatable amounts of RAM memory.

Considering a big European airspace of 5000Km2 and 20 flight levels (6000m), and

applying the predictor system’s concept that uses a granularity of 100 meters to have a

better resolution of the European air traffic scenario, the memory requirements would

be:

The necessary amount of memory to store only the BS database is impossible to support

by the current commercial computers, thus the wake vortex encounter predictor system

is considerably better in terms of resolution and in computational efficiency, being more

effective and safer for the strategic management purposes.

To evaluate the proposed system’s efficiency in terms of memory requirements, we

have to consider first the memory required for performing the HRRI pre-process. Since

the configuration of the SDS is based on RTSDS and similar configuration than the one

seen in section 2.3.2.3 (RTSDS: combination of RSDS and TSDS), the maximum

amount of memory required for the HRRI module is 45GB approximately. To this

quantity we have to add the memory requirements of the WV Risk Analyzer, which is

dependent on the number of HRR regions to be explored and the size of these.

(15)

(16)

(17)

(18)

X = Y = 5000000m/100m = 50.000

Z = 6000m/100m = 60

T = 5h · 3600s/h = 18000

BS= X·Y·Z·T·4 = 50.000 · 50.000 · 60 · 18000 · 4 = 10.800 TB

STI = N · L · 8 = 30000 · 2 · 3600 · 8 = 1.8GB (19)


- 47 -

The number of 2000 HRR will be considered as the maximum based on the

approximation obtained through simulations using conflict detection technology and

considering the absence of flight level constraints, the freedom of flight route and

10NM minimum separation to identify the HRR.

With regards the size of the HRRs, a deeper knowledge is necessary to determine with

precision which can be the area of risk of a wake vortex generated. For this reason will

perform calculations for the worst-case and the most-likely scenarios regarding the

memory needs to store the entire state-space of the European airspace at micro-scale.

The worst-case scenario is considered to be a conservative configuration of the system

that leads to the identification of larger areas of risk (i.e., HRR) that need to be zoomed

in with a microscopic representation of the vortex and trajectories, thus requiring a lot

of memory positions for their representation. In this case, the highly conservative HRR

regions are considered to be of size 40NM x 40NM x 4000ft.

The most-likely scenario is considered to lead to conservative HRR sizes but much

closer to the typical values of current safety distances, i.e., 20NM x 20NM x 2000ft, and

therefore expected to be more realistic.

4.1 Most-likely Scenario

First of all we have to calculate how many cells there will be, knowing that the

corresponding HRR’s granularity is 100 meters.

Horizontally:

20 NM = 37.040 m

37.040m/100 m = 370,4 cells ~ 371 cells

Vertically:

2000ft = 610 m

610m/100 m = 6,1 cells ~ 7 cells

Therefore, the total number of cells is obtained multiplying the number of cells in

each axis:

Number of cells for each HRR = 371 x 371 x 7 = 963.487 cells

Then, since each cell occupies 4 bytes, and the expected number of HRR is 2000,

the necessary amount of memory to store the BS will be:


- 48 -

BS = 963.487 x 4 x 2000 = 7, 8 GB

A part from that, we have to calculate the memory occupied by the WVI (Wake

Vortex Information) which is substituting the role of the STI in the CD process,

using the following equation:

where TM refers to the time of simulation, #points to the number of point positions

(i.e. coordinates) considered to configure the WV’s discrete model and N to the

number of trajectories that we have to model.

So, in this case, we will take into account the worst-case scenario in order to see if

the system is capable to support the storage of all these information. Therefore, the

number of trajectory's seconds to model for each region will be 3600, while amount

of point positions will be 10 and the amount of trajectories 4000, obtaining the next

result:

WVI = 3600 x 10 x 4 x 4000 = 0, 6 GB

However, as it is mentioned previously, this is the worst-case scenario. Thus, the

expected one would occupy approximately a 5% of the result (i.e. 29 MB), since it

depends on the number of regions, which is obtained through the interaction between

trajectories, and the regions’ size, which determines how many seconds of the

trajectory we have to model for each region, obtaining values of the TM and N

considerably lower. This result will be also used in the worst case, since the

reasoning is the same.

So that, the total amount of memory required when the expected case is

contemplated is:

Total Amount of Memory Required: BS + WVI = 7,8 GB + 0,6 GB = 8,4 GB

WVI = TM x # points x 4 x N (21)

(20)

(22)

(23)


- 49 -

4.2 Worst case scenario

The same process is repeated for the worst-case scenario, i.e. (40NM x 40NM x

4000ft) whose objective is to check the system’s efficiency even considering very

conservative distances.

Horizontally:

40 NM = 74.080 m

74.080 m/100 m = 740,8 cells ~ 741 cells

Vertically:

4000ft = 1220 m

1220 m/100 m = 12,20 cells ~ 13 cells

Therefore, the total number of cells is obtained multiplying the number of cells in

each axis:

Number of cells for each HRR = 741 x 741 x 13 = 7.138.053 cells

Then, since each cell occupies 4 bytes, and the expected number of HRR is 2000,

the necessary amount of memory to store the BS will be:

BS = 7.138.053 x 4 x 2000 = 57, 2 GB

Moreover, we have to add the amount of memory occupied by the WVI (Wake

Vortex Information) as well as it was done in the expected case:

WVI = 3600 x 10 x 4 x 4000 = 0, 6 GB

So that, the total amount of memory required when the worst case is contemplated is:

Total Amount of Memory Required: BS + WVI = 57,2 GB + 57,8 GB = 57,8 GB

(24)

(25)

(26)


- 50 -

As we can observe the amount of memory in both the most-likely and the worst-case

scenarios, together with the needs of memory of the HRRI module, is easily supportable

by the capacity of current computers, which can accept up to 2TB with quasi-standard

configurations, so the main objective has been achieved for both expected case and the

worst case in which we are considering large scenarios.

Conclusions

- 51 -

5. CONCLUSIONS

The forecasts indicate a notable growth of the air traffic in the coming years, which

implies more airspace congestion and, consequently, the necessity to carry out a more

efficient management aimed at achieving an airspace free of conflicts and free of wake

vortex encounter hazards.

The potential consequences of wake vortex encounters make it a phenomenon that

requires special attention. For this reason, safety distances were defined to reduce the

risk of encounter. However, nowadays, and considering the characteristics of current

and future ATM, these distances are not enough, since it is a phenomenon whose

behaviour and evolution in time and space depends on several factors, among which the

meteorological conditions that brings a lot of uncertainty into the forecast, or the actual

aircraft mass during the flight.

Thus, it is necessary to have a system able to predict with enough precision when and

where a wake vortex encounter may occur, and quantify with accuracy the severity

associated. The system architecture proposed in this project could meet with all these

requirements with enough efficiency to process thousands of trajectories flying in the

European airspace for look-ahead times of several hours.

In this sense, the main contributions of the proposed system have been to use the

technology of Spatial Data Structures to identify regions with a high probability of wake

vortex encounter, to filter and reduce the regions that are subject to a thorough micro-

scale analysis of the evolution of the wake vortex phenomenon. Spatial Data Structures

have been also proposed to support the efficient prediction of wake vortex encounters,

to quantify their severity and to determine the affected aircraft.

The proposed method is based on the combined use of different pre-existing SDS

configurations that have been enhanced and integrated to take full advantage of the

strengths and diminish the shortcomings of each configuration. The resulting system

shall allow to be implemented in a regular commercial computer to process large set of

traffics at European airspace scale in timely manner, a problem that is not trivial due to

the large amount of memory and/or time required.

Conclusions

- 52 -

In addition, in this project a better vortex's discretization has been proposed with respect

previous research, achieving a more precise representation of the wake vortex

dynamics. In previous works the tube built to model vortex and to perform wake vortex

encounter detection had 300 meter of radius (1000ft) to take into account the worst-case

dynamics of wake vortex, but in this system, with the discretization and model carried

out it is possible to know the position of the vortex and strength for each discrete time

step with much more precision.

Therefore, with the proposed system and algorithms it shall be possible to predict all the

potential of wake vortex encounter and to predict the resulting consequences, thus

enhancing the traffic planning processes and ATM decision-making processes with a

system that can contribute to the current ATM and also fit perfectly into the SESAR's

proposed ATM.

Bibliography

- 53 -

6. BIBLIOGRAPHY

[1] Ruiz, S., Piera, M.A., 2013. Relational time-space data structure to enable

strategic de-confliction with a global scope in presence of large number of 4D

trajectories. Journal of Aerospace Operations, 2, 53-78.

[2] Ruiz, S., Piera, M.A., del Pozo, I., 2012. A Medium Term Conflict Detection

and Resolution system for Terminal Maneuvering Area based on Spatial Data

Structures and 4D Trajectories. Transportation Research Part C, 26, 396-417.

[3] Ruiz, S., Piera, M.A., Nosedal, J., Rainieri, A., 2013. Strategic de-confliction in

the presence of a large number of 4D trajectories using a causal modeling

approach. Transportation Research Part C, 39, 129-147.

[4] DE CLEYN, K., 2011. Forecast based Wake Turbulence Separation. Master of

Science Final Thesis Report. Delft University of Technology.

EUROCONTROL.

[5] Ruiz, S., 2010. Sistema de Detección y Resolución de Conflictos basado en

Estructuras de Datos Espaciales. Universitat Autònoma de Barcelona.

[6] Ruiz, S., 2013. Strategic Trajectory De-confliction to Enable Seamless Aircraft

Conflict Management. PhD Thesis. Universitat Autònoma de Barcelona.

[7] EUROCONTROL. Single European Sky, 2015.

https://www.eurocontrol.int/dossiers/single-european-sky.

[8] SESAR Joint Undertaking. Background on Single European Sky, 2015.

http://www.sesarju.eu/discover-sesar/history/background-ses.

[9] European Comission. Transport Modes. Single European Sky, 2015.

http://ec.europa.eu/transport/modes/air/single_european_sky/.

[10] SESAR Joint Undertaking. Discover SESAR. Objectives, 2015.

http://www.sesarju.eu/discover-sesar/objectives //.

https://www.eurocontrol.int/dossiers/single-european-sky

http://www.sesarju.eu/discover-sesar/history/background-ses

http://ec.europa.eu/transport/modes/air/single_european_sky/

http://www.sesarju.eu/discover-sesar/objectives%20/

Bibliography

- 54 -

[11] Single European Sky ATM Research. SESAR in brief, 2008.

http://www.euro-cdm.org/library/sesar_in_brief_2008_final_low.pdf.

[12] CDTI. SESAR. El futuro de la Gestión del Tránsito Aéreo en Europa: El Cielo Único

Europeo y la JU SESAR.

https://www.cdti.es/recursos/doc/Programas/Cooperacion_internacional/P.Marco_I%20

D_de_la_UE/Transporte/31200_21112111200894122.pdf.

[13] EUROCONTROL. Wake Vortex, 2015.

https://www.eurocontrol.int/articles/wake-vortex.

[14] SKYbrary. En-route Wake Vortex Hazard, 2013.

http://skybrary.aero/index.php/En-route_Wake_Vortex_Hazard.

[15] Good Aviation Practice. Civil Aviation Authority of New Zeland. Wake Turbulence,

2008. file:///C:/Users/User/Downloads/Wake_Turbulence.pdf.

[16] FAA. Aircraft Wake Turbulence, 2002.

http://www.faa.gov/documentLibrary/media/Advisory_Circular/AC90-23f.pdf.

[17] PilotFriend. Flight Safety. Wake turbulence.

http://www.pilotfriend.com/safe/safety/wake_turb.htm.

[18] WETTER & FLIEGEN. DLR project. Wake Vortex Physics and Encounter

Consequences. http://elib.dlr.de/76397/1/holz-1.pdf.

[19] Klaus-Uwe Hahn. Institute of Flight Research, German Aerospace Center (DLR). ICAS

2002 CONGRESS. Coping with Wake Vortex. http://wakenet.eu/fileadmin/wakenet2-

europe/Wakenet2/Publications/Publications,%20Papers,%20Conferences/ICA0250P.pdf.

[20] Avstop. Section 3. Wake Turbulence. http://avstop.com/ac/aim/chap7/aim0703.html.

[21] SESAR. The SESAR Programme R&I 2020 future developments.

http://www.iserd.org.il/_Uploads/dbsAttachedFiles/JU_SESAR_2020_v2.pdf

[22] ATM Seminar Severity Metric, 2015.

http://www.atmseminar.org/seminarContent/seminar11/papers/425-De%20Visscher_0126151141-

Final-Paper-5-5-15.pdf

http://www.euro-cdm.org/library/sesar_in_brief_2008_final_low.pdf

https://www.cdti.es/recursos/doc/Programas/Cooperacion_internacional/P.Marco_I

https://www.cdti.es/recursos/doc/Programas/Cooperacion_internacional/P.Marco_I

https://www.eurocontrol.int/articles/wake-vortex

http://skybrary.aero/index.php/En-route_Wake_Vortex_Hazard

http://www.faa.gov/documentLibrary/media/Advisory_Circular/AC90-23f.pdf

http://www.pilotfriend.com/safe/safety/wake_turb.htm

http://wakenet.eu/fileadmin/wakenet2-

http://wakenet.eu/fileadmin/wakenet2-

http://avstop.com/ac/aim/chap7/aim0703.html

http://www.iserd.org.il/_Uploads/dbsAttachedFiles/JU_SESAR_2020_v2.pdf

A NEW WAKE VORTEX ENCOUNTER PREDICTION … · de encuentros con estelas turbulentas, a fin de...

Documents

Transcript of A NEW WAKE VORTEX ENCOUNTER PREDICTION … · de encuentros con estelas turbulentas, a fin de...